Khảo sát kỹ thuật học sâu trên bài toán chẩn đoán hư hỏng động cơ điện dựa trên tiếng ồn vận hành
Abstract
Early detection of induction motor failure plays an important role in limiting disruption to industrial production. Sensor-based measurement methods are highly reliable, but the installation of the equipment is time consuming and costly. Building a smartphone’s application to diagnose electric motor problems is a research direction that attracts many groups. This paper proposes to investigate the ability to train and diagnose electric motor faults based on the principle of recognizing scalogram images of motor’s operation sounds, using a deep learning neural network. The audio signals are noise-filtered, amplitude normalized, and scalogram rendering by wavelet transforms. The set of scalogram images is divided into two parts for training and validating the GoogLeNet convolutional neural network. The GoogLeNet is also investigated through changing some basic parameters, in order to determine the best training efficiency. After training, the network is tested on an independent sound signal dataset. The results show that the network is able to identify 3 common motor problems including phase loss, insulating film brush and bearing failure with 94.21% accuracy. The experiment also shows that the development of smartphone’s application for early diagnosing electric motor problems is feasible.
Tóm tắt
Phát hiện sớm sự cố động cơ điện góp phần hạn chế gián đoạn hoạt động sản xuất công nghiệp. Phương pháp đo dùng cảm biến có độ tin cậy cao, song việc lắp đặt mất thời gian và chi phí. Việc xây dựng ứng dụng điện thoại để chẩn đoán sự cố động cơ điện thu hút nhiều nghiên cứu. Bài báo tiến hành khảo sát khả năng chẩn đoán lỗi động cơ điện thông qua nhận diện ảnh phổ tín hiệu âm thanh vận hành dùng mạng neuron học sâu GoogLeNet. Dữ liệu âm thanh được lọc nhiễu, chuẩn hóa biên độ và dựng ảnh phổ bằng phép biến đổi wavelet. Tập ảnh phổ được dùng để huấn luyện và kiểm tra mạng. Mạng GoogLeNet cũng được khảo sát hiệu quả huấn luyện thông qua việc thay đổi các tham số cơ bản. Sau đó, mạng được kiểm tra trên tập dữ liệu độc lập. Kết quả cho thấy mạng nhận diện 3 sự cố thông dụng, gồm mất pha, cọ phim và hỏng bạc đạn, với tỷ lệ chính xác đạt 94,21%. Thí nghiệm cũng cho thấy khả năng phát triển ứng dụng điện thoại là khả thi.
Article Details
Tài liệu tham khảo
Akansu, A. N. (1994). Wavelets and filter banks A signal processing perspective. IEEE Circuits and Devices Magazine, 10, 14-18.
Arora, S., Bhaskara, A., Ge, R., & Ma, T. (2014). Provable bounds for learning some deep representations. Proc. of the 31st Inter. Confer. on Machine Learning, in PMLR, 32(1), 584-592.
Bonnett, A.H., & Soukup, G.C. (1992). Cause and analysis of stator and rotor failures in three-phase squirrel-cage induction motors. IEEE Transactions on Industry Applications, 28(4), 921-937. DOI: 10.1109/28.148460.
Daubechies, I. (1992). Ten Lectures on Wavelets. CBMS-NSF Regional Conference Series in Applied Mathematics, Philadelphia, SIAM, 1st Ed. ISBN: 978-0-89871-274-2.
Chen, Y.-W. (2021). Audio Normalization. MATLAB Central File Exchange (retrieved 10/8/2021).
Ciszewski, T., Gelman, L., & Swedrowski, L. (2016). Current-based higher-order spectral covariance as a bearing diagnostic feature for induction motors. Insight - Non-Destructive Testing and Condition Monitoring, 58(8), 431-434.
Glowacz, A., & Glowacz, Z. (2017). Diagnosis of the three-phase induction motor using thermal imaging. Infrared Physics & Technology, 81, 7-16, https://doi.org/10.1016/j.infrared.2016.12.003.
Glowacz, A., Glowacz, W., Glowacz, Z., Kozik, J., Gutten, M., Korenciak, D., Khan, Z.F., Irfan, M., & Carletti, E. (2017). Fault diagnosis of three phase induction motor using current signal, MSAF-Ratio15 and selected classifiers. Arch Metall Mater, 62(4), 2413–2419.
Glowacz, A., Glowacz, W., Glowacz, Z., & Kozik, J. (2018). Early fault diagnosis of bearing and stator faults of the single-phase induction motor using acoustic signals. Measurement, 113, 1-9. http://dx.doi.org/10.1016/j.measurement.2017.08.036.
Guo, P., Infield, D., & Yang, X. (2012). Wind turbine generator condition monitoring using temperature trend analysis. IEEE Trans. Sustain. Energy, 3(1),124–133.
Haar, A. (2010). Zur Theorie der orthogonalen Funktionensysteme. Mathematische Annalen, 69(3), 331–371. DOI:10.1007/BF01456326.
Henao, H., Capolino, G.A., Fernandez-Cabanas, M., Filippetti, F., Bruzzese, C. , Strangas, E., Pusca, R., Estima, I., Riera-Guasp, M., & Hedayati-Kia, S. (2014). Trends in Fault Diagnosis for Electrical Machines A Review of Diagnostic Techniques. IEEE Industrial Electronics Magazine, 8(2), 31-42. DOI:10.1109/MIE.2013.2287651.
Ince, T., Kiranyaz, S., Eren, L., Askar, M., & Gabbouj, M. (2016). Real-Time Motor Fault Detection by 1-D Convolutional Neural Networks. IEEE Transactions on Industrial Electronics, 63(11), 7067-7075. DOI:10.1109/TIE.2016.2582729.
Kim, D., Kim, H., Hong, J., & Park, C. (2014). Estimation of Acoustic Noise and Vibration in an Induction Machine Considering Rotor Eccentricity. IEEE Transactions on Magnetics, 50(2), 857-860. DOI:10.1109/TMAG.2013.2285391.
Kumar, A., Gandhi, C.P., Zhou, Y., Kumar, R., & Xiang, J. (2020). Improved deep convolution neural network (CNN) for the identification of defects in the centrifugal pump using acoustic images. Applied Acoustics, 167, 107399. ISSN 0003-682X. https://doi.org/10.1016/j.apacoust.2020.107399.
Lilly, J.M., & Olhede, S. C. (2020). Generalized Morse Wavelets as a Superfamily of Analytic Wavelets. IEEE Transactions on Signal Processing, 60(11), 6036-6041. DOI:10.1109/TSP.2012.2210890.
Li, Y., Chai, F., Song, Z., & Li, Z. (2017). Analysis of Vibrations in Interior Permanent Magnet Synchronous Motors Considering Air-Gap Deformation. Energies, 10(9), 1259. https://doi.org/10.3390/en10091259.
Mallat, S. G. (1989). A Theory for Multiresolution Signal Decomposition: The Wavelet Representation. IEEE Trans. on Pattern Analysis and Machine Intelligence, 11(7), 674–693.
Mishra, A., Ranjan, P., & Ujlayan, A. (2020). Empirical analysis of deep learning networks for affective video tagging. Multimedia Tools Applications, 79, 18611–18626, Springer. https://doi.org/10.1007/s11042-020-08714-y
Qi, Y., Shen, C., Wang, D., Shi, J., Jiang, X., & Zhu, Z. (2017). Stacked Sparse Auto encoder-Based Deep Network for Fault Diagnosis of Rotating Machinery. IEEE Access, 5, 15066–15079.
Sangeetha, P., & Hemamalini, S. (2017). Dyadic wavelet transform-based acoustic signal analysis for torque prediction of a three-phase induction motor. IET Signal Processing, 11(5), 604–612. DOI: 10.1049/iet-spr.2016.0165.
Singh, S.K., Kumar, S., & Dwivedi, J.P. (2017). Compound fault prediction of rolling bearing using multimedia data. Multimedia Tools Applications, 76, 18771–18788. https://doi.org/10.1007/s11042-017-4419-1.
Szegedy, C., Wei, L., Yangqing, J., Sermanet, P., Reed, S., Anguelov, D., Erhan, D., Vanhoucke, V., & Rabinovich, A. (2015). Going Deeper with Convolutions. IEEE Confer. on Computer Vision and Pattern Recognition – CVPR 2015 (pp. 1-9). DOI: 10.1109/CVPR.2015.7298594.
Wen, L., Gao, L., & Li, X. (2017). A New Deep Transfer Learning Based on Sparse Auto-Encoder for Fault Diagnosis. IEEE Transactions on Systems, Man, and Cybernetics: Systems, 49(1), 136-144. DOI:10.1109/TSMC.2017.2754287.
Xia, M., Li, T., Xu, L., Liu, L., & de Silva, C.W. (2018). Fault Diagnosis for Rotating Machinery Using Multiple Sensors and Convolutional Neural Networks. IEEE/ASME Transactions on Mechatronics, 23(1), 101-110. DOI:10.1109/TMECH.2017.2728371.
Yang, T., Pen, H., Wang, Z., & Chang, C.S. (2016). Feature Knowledge Based Fault Detection of Induction Motors Through the Analysis of Stator Current Data. IEEE Transactions on Instrumentation and Measurement, 65(3), 549-558. DOI: 10.1109/TIM.2015.2498978.
Zhou, B., Khosla, A., Lapedriza, A., Torralba, A., & Oliva, A. (2014). Places: An image database for deep scene understanding. Journal of Vision, 17(10), 296, 1-12. DOI: https://doi.org/10.1167/17.10.296.